THE BOOK--Playing The Percentages In Baseball

Only explanation I can think of: Black thought or hoped he could get 2 more innings out of Peavy, but then Peavy threw 27 pitches in the 7th to get to 116.

"Why did Peavey bat in the 6th?”

Because in modern baseball, if you have a star pitcher like Peavey, you let him stay in to reach his normal pitch count, if he is near his usual effectiveness, and the game is close.

Right or wrong, I don’t know. It all depends on what you include in the analysis. My instinct is that the typical sabermetric analysis is lacking, considering that these trends have been only increasing, as time goes on.

How is the traditional sabermetric analysis lacking in this case?  It is fairly clear cut.  What is the win expectancy in that situation with Peavy batting, versus what is the WE with a pinch hitter and another pitcher for 1-2 innings?  And of course you have to consider “externalities” to use an economic term, such as the long-term effect on your bullpen, one less pinch hitter for the rest of the game, and the possible long-term effect on Peavy for being taken out of a game in which he is pitching well (although I personally don’t think getting yanked in the 6th inning of a close game with the bases loaded and 1 out is anything to get upset about).

My general rule is that if a decision is close, then you can give the benefit of the doubt to the “externalities” in favor of whatever the manager wants to do or whatever conventional wisdom says to do.  If the decision is not even close, which I don’t think it is in this situation, then all the externalities in the world are not going to change the correct answer.

The Padres bullpen, prior to last night, had thrown 30 innings in their previous 7 games and about 45 or so over their previous 10.  That includes a pair of 13 inning games and the 22 inning game. 

I’m not saying that justifies the decision Black made, but it surely had some impact.

So, it seems like a closer decision now—and I’d like to say I like this question a lot.

I assume the manager goes over his list of relievers before every game and puts them on a spectrum that looks like this: he should pitch, he can pitch, he can pitch to one hitter, he shouldn’t pitch but could in a pinch, he really needs the day off.

Based on #4 David’s post, Black had a lot of guys in the lower categories. So maybe it was a season-level decision even though it wasn’t the optimal game-level decision.

This is interesting, as back in the 50s and 60s, Koufax and other pitchers had LOTS of games where his pitch count was well below 100 pitches.  I can only guess this was because a manager brought in a pinch hitter for him.  We can look more closely at the game logs.

David/4: let’s talk in terms of days, not games.  And it should be pitches, or lacking that, batters faced.

The average bullpen will probably throw 500 pitches over 10 days (say face about 130 batters).  Going 10%, even 20% above that in any 10 day period is probably normal.  In any case, I wouldn’t look past 5 or 6 days anyway.

What were the Padres like?

Also note that if their pitchers have options, they can bring guys up/down, especially if guys are used up.

Having said all that, the LI was 4.23:
http://www.fangraphs.com/plays.aspx?date=2008-04-27&team=Padres&dh=0

I’m guessing the difference between him and a pinch hitter would have been about .07 wins.

The difference between a guy with a 3.00 runs per game allowed and one with 6.00 runs per game allowed is 3 runs per 9 IP, or 0.3 wins per 9 IP or .03 wins per IP.  You could have replaced him with a pitcher allowing 9 runs per game, and you’d still be ahead.

They could have gone to a starter on his throw day.  They could have gone to whatever reliever they really don’t care about.  What’s wrong with Thatcher here?  It looks like he faced 4 batters the day before, had an off day, and 8 before that.  In terms of his “Tank”, he faced 8 “effective” batters.  A bit high, but not unusually high.

I am glad that Tango better quantified the amount lost by not pinch hitting in that situation.

You can give me all the reasons why Black decided not to pinch hit for Peavy, but I stand by my statement that when the difference is not even close, all the “other factors” in the world are not likely to change the answer.

Tango hit the nail on the head when he said that they could have put the bat boy in there to pitch (not literally of course) and would still be “ahead” with the pinch hit.  If you don’t really care about the quality of the pitcher, in terms of your decision, I don’t see how you can be overly concerned with taxing your pen. All teams, should at least, have back of the bullpen relievers who are there only to pitch in low leverage situations and are completely fungible (such that if you use them up, you simply go and get another one from the minors).  Even though the 7th inning would not likely be a low leverage situation (of course it had more of a chance to be low leverage - the Pads breaking the game open - if you put in a pinch hitter), as Tango pointed out, even if you put the worst reliever on the planet in for the 7th inning, you would still be ahead of the game.

My general rule is a good one I think, and is exactly the way I approach almost all of my analyses.  That is, I first look at the difference in WE or RE between two (or more) alternative strategies, without necessarily getting too thorough in the analysis. If that difference is large, in most cases it is not necessary to consider all possible factors, since they likely won’t change the answer.

We can usually put a reasonable “cap” on how much of an effect an “externality” (I love that word) is likely to have.  Anyway, if the decision seems close, after a non-thorough analysis, then we can either do a better job with the analysis and come up with an updated answer (albeit with a lower level of confidence or reliability, if the difference still remains close), or we can declare it a coin flip, such as Tango did with his “IBB Bonds” chart.

It’s also possible that Black just blew it. Seems like a question worth asking the manager after the game, doesn’t it? As far as I can tell, no one did.

I think one of the reasons why Managers let a (good) starter bat in this type of situation and continue pitching for an inning is to give him a chance for a “Win” (or at least to let him avoid the “Loss").

Not that makes any sort of logical sense… But I still think it factors into the descision. One could also debate where the managers loyalty lies here, both in practice and how it “should” be.

I don’t think this is a case where the manager would say, “Yeah, I blew it, I made the wrong decision, next time I will never let my pitcher bat in that situation, even my best pitcher, and one of the best pitchers in baseball.” No.  Definitely not.

This is a typical case whereby the manager has NO IDEA that he is costing his team win expectancy.  How would he know that?  Managers do what managers have been doing, more or less, for a 100 years.  Sometimes manager tendencies change, and then managers do what managers have been doing for 5 or 10 years.  That’s all.  How is a manager to know what are the “correct” strategies, unless they read books like The Book and the like?  None of them do.  And if they do, none of them implement the ideas.  None. (Yes, before anyone’s head explodes, managers do a lot of correct things - not all things they do are incorrect. And yes, it is possible to actually come up with the “correct strategies” for some things and in some instances, just from experience, and whatever “wisdom” they do possess.)

I would wager a million dollars that if you asked Black if leaving Peavy in gave him a better or worse chance of winning, his answer, without hesitation, would be, “Better of course.  What do you think, I am an idiot?  I would do something that would give my team a LESSER chance of winning?  Why would I do that?”

He would NOT say, “Yes I know that we significantly reduced our chances of winning that game, even as compared to replacing him with my worst reliever, but I wanted to leave Peavy in the game to save my bullpen for another inning or two.” No way he is thinking that. 

There are always “reasons” for doing things.  That does not make the “thing” correct or the “reasons” rational or correct or grounded in fact, data, or evidence.

I am pretty sure his “reasons” were that Peavy was pitching well, he is a great pitcher, had another 1-2 innings in him, is a pretty good hitter for a pitcher, there was only one out, it was a close game, and he wanted Peavy to stay and try and get the win (which is a little silly, since if he pinch hits for him and they score 2 runs, he gets the win anyway, so he probably has a better chance to get the win if he comes out!).

All of those “reasons” are true and valid and make sense to Black and to most people in general, but it does not change the fact that he likely greatly reduced his chances of winning the game by letting him bat, and more importantly, that he made his decision with the wrong information - he thought that leaving him in there would increase his chances of winning rather than decrease them - I am presuming.

What would it take for the team to provide someone whom he trusts and believes to take him aside after the game and just explain to him the facts, and then tell him he is still allowed to make his own decision; they just want him to have all the relevant information?  That doesn’t seem like a tall order to me.

As I said, I would wager a million bucks that he has NO idea that letting him bat significantly reduces his team’s winning percentage.  How would he know that, and if he DID know that, he would likely (not definitely) have taken Peavy out.

That is a common mistake among managers in general - letting a pitcher bat in a high leverage situation in the 4-7th innings.  It is particularly irksome to me when the pitcher only pitches another inning or less after the AB, which happens some of the time.

In the history of Retrosheet, this situation:

bases loaded
1 out
6th inning
home team batting
9th slot up

has occurred 620 times.  The pitcher came to bat 212 times.

If I include a score of the home team tied, down by 1 or 2, then: 29 out of 177.  So, only 16% of the time has a manager faced with a Peavy situation allowed his pitcher to bat.

I did not, yet, look to see the quality of the pitcher, nor how many batters the pitcher had already faced.  If someone wants to do so:
http://tangotiger.net/files/peavy_ph.html

Hmmm… I should have limited that list to NL parks only, obviously.

I’d like to know: Who won the games?

I also wonder if it’s possible to put some weight on the games themselves. If the most important regular season game is a 163rd-game playoff and ranks in importance at 100 on a scale of 100, where do you rank a game on April 27 ... against the division-leading team ... that’s 5 games ahead of you?

There has to be a formula for calculating game importance, doesn’t there?